Data from 35 trials with 482 lactating cows fed 106 diets were used to study the effects of animal and dietary factors on the relationship between milk and blood urea N and the value of milk urea N in the assessment of protein status. In two trials, urea N in whole blood and in blood plasma were closely related (r2 = 0.952); the slope was not significantly different from 1.0, and the intercept was not significantly different from 0. Regression of milk urea N on blood urea N with a mixed effects model using all 2231 observations yielded the equation: milk urea N (milligrams of N per deciliter) = 0.620 x blood urea N (milligrams of N per deciliter) + 4.75 (r2 = 0.842); this model accounted for a significant interaction of cow and blood urea N. Single factors that yielded significant regressions on milk urea N with the mixed effects models were dietary crude protein (CP) (percentage of dry matter; r2 = 0.839), dietary CP per megacalorie of net energy for lactation (NEL) (r2 = 0.833), excess N intake (r2 = 0.772), N efficiency (r2 = 0.626), and ruminal NH3 (r2 = 0.574). When all factors were analyzed at once, 12 were significant in a mixed effects model. Blood urea N, body weight, yield of fat-corrected milk, dietary CP content, excess N intake, dry matter intake, and days in milk were positively related to milk urea N, and parity, milk and fat yield, dietary CP per unit of NEL content, and NEL intake were negatively related to milk urea N. In one trial, the mean urea concentration was 35 times greater in urine than in milk; lower proportions of total urea excretion in milk were observed in the a.m. sampling (1.8%) than in the p.m. sampling (3.3%). Measuring urea N in a composite milk sample from the whole day substantially improved reliability of data. The number of cows fed a specific diet that must be sampled to determine mean milk urea N within 95% confidence intervals with half widths of 1.0 and 2.0 mg of N/dl was estimated to be 16 and 4, respectively.